Skip to main content
Each page is a request you can drop into an app. You can launch most pipelines we list, then call them the same way: matching task, API key, SYNAPSAI_API_KEY, and your-model-id from Models.
Prototype in the Playground, then keep the same body in Python, cURL, or an OpenAI-compatible integration.

​
Inference

Text generation

Chat completions, completions, streaming.

Responses API

Create, stream, store, and continue a response.

Tool calling

Function calling on chat completions.

Embeddings

Vector embeddings for text.

Images

Generation, edits, detection, segmentation.

Audio

Speech, transcription, translation, generate.

Videos

Image-text-to-video jobs.

Classification

Text, image, audio, video, zero-shot.

Question answering

Extractive, document, table, and visual QA.

Rerank

Rank documents for a query.

Fill mask

Masked language modeling.

Feature extraction

Hidden-state feature vectors.

​
Product APIs

Agents

Run a saved agent (AG-UI stream) from your backend.

Knowledge bases

Vector store search and file upload.

Integrations

LangChain, LlamaIndex, FAISS, Gradio, and FastAPI using the same deployment id.